The Research On Multi-VM Fast Deployment Mechanism

The rapid development of the virtualization technology has provided a good opportunity to the efficient use of computing resources. Multi-VM application management has become a hot research in virtualization field. However, the management of Multi-VM is still in the preliminary stage, how to efficiently manage and construct user's task environments remains a significant challenge.Multi-VM rapid Deploy System provides an application deployment platform based on the VMM and Cluster environment, it achieves automatic deployment of Multi-VM by using template and incremental mechanism, provides a theoretical basis for the virtualization technology research. The architecture of Multi-VM rapid Deploy System has three layers: User Interface layer presents functional interface through portal, including the template list, VM list and VM deployment portal, users can call the related services through these interfaces; Strategy Service layer is located in the front-end machine, it makes response to the operation of the User Interface layer, the main services provided include template management, template scheduling and VM deployment, template management is responsible for the definition and storage of the templates, as well as the relationship between the templates and the users, template scheduling refers to template pre-copy strategy, template deployment is the strategies and technologies about deployment, such as deployment strategy, incremental and parallel mechanism; Control Service layer mainly provides the concrete realization of the deployment service, including extracting information of the servers, the creation and launch of the VM. In this layered architecture, each layer has its own services and interfaces which are invoked by its upper layer, layers are flexible and independent between each other. This makes the system easy for implementation and maintenance.The results of function tests indicate that Multi-VM rapid Deploy System is full functional and works normally. The results of performance tests illustrate that the deployment response time will reduce 85% when incremental strategy is added and the response time will reduce 44.5% after adding parallel policy, many virtual machines can be deployed in a few minutes.